Development of Partial Least Squares Integrated Fourier Transform Infrared Approach for Prediction of Moisture Content in Transformer Oil and Lubricating Oil

Amelia Laccy, Jeffrey Kimura (2021) Development of Partial Least Squares Integrated Fourier Transform Infrared Approach for Prediction of Moisture Content in Transformer Oil and Lubricating Oil. Masters thesis, Universiti Malaysia Sarawak.

[img] PDF (Please get the password from Digital Collection Development Unit, ext : 3913 / 3914)
Amelia Laccy.pdf
Restricted to Registered users only

Download (1MB) | Request a copy

Abstract

Monitoring of moisture content in oil by utility companies is a routine exercise to keep the oil performance in check. Currently, Karl Fischer titration is being used for water determination in oil with sensitivity of 10 ppm and below however, it involves various expensive solvents and its time-consuming. Similarly, the Kittiwake analyzer also requires expensive solvents for analysis. In this study, Fourier Transform Infrared (FTIR) incorporated with Partial Least Squares was studied for prediction of moisture in the locally available transformer oil and lubricating oil. The standards were prepared by direct spiking of moisture in oil, direct spiking of moisture with addition of surfactant to reduce scattering of infrared light and extraction of moisture using acetonitrile. Among the three strategies, the most effective way to prepare standards for FTIR is the solvent extraction method. The spectral regions corresponding to moisture were found at 1600-1700 cm-1 and 3400-3750 cm-1. The former region at the lower frequency was found to produce better prediction accuracy with a lower % Root Mean Squares Error (RMSE). The FTIR method incorporated with PLS regression predicts samples with higher moisture concentrations with better accuracy. The method is not sensitive for detection of moisture at concentrations less than 1000 ppm. The limit of detection (LOD) established for quantification of moisture in transformer oil and lubricating oil was 1000 ppm and 700 ppm, respectively. The low sensitivity of FTIR is attributable to the short path length cell. The Karl Fischer (KF) method and Kittiwake analyser were used to compare with the FTIR method. However, both methods were not directly comparable with the FTIR method because both KF and Kittiwake demonstrated a better sensitivity for detection of moisture at low concentrations.

Item Type: Thesis (Masters)
Additional Information: Thesis (MSc.) - Universiti Malaysia Sarawak , 2021.
Uncontrolled Keywords: Partial Least Squares regression, training/test sets, Karl Fischer, Kittiwake, Percentage Root Mean Squares Error.
Subjects: Q Science > QD Chemistry
Divisions: Academic Faculties, Institutes and Centres > Faculty of Resource Science and Technology
Depositing User: AMELIA LACCY ANAK JEFFREY KIMURA
Date Deposited: 17 Oct 2021 03:01
Last Modified: 22 Oct 2021 01:00
URI: http://ir.unimas.my/id/eprint/36436

Actions (For repository members only: login required)

View Item View Item